symptoms and threat of disease, (4) social network characteristics, ... since one might expect each model builder to wish to preserve what are felt to represent the ...
Journal of BehavioralMedicine, Vol. 3, No. 2, 1980
Bringing the Models Together: An Empirical Approach to Combining Variables Used to Explain Health Actions K. Michael Cummings, 1 Marshall H. Becker, 1 and Maria C. Maile I
Accepted for publication. January 11, 1980
Considerable confusion has existed among researchers with regard to the selection o f a particular model o f health behavior for study, and many investigators have long felt that the actual number o f truly distinct concepts relevant to explaining health-related actions is considerably lower than the large number o f variables currently employed. This paper explores selected approaches and models which have been advanced to explain health actions, in terms of structural similarities and differences identified by a panel of/udges who are the relevant experts in this field. Judges were asked to partition a set o f 109 variables, representing 14 different models, into 1 2 - 1 4 groups on the basis o f similarity. The structural similarities among the variable groups were evaluated using Smallest Space Analysis. Six interpretable factors emerged from the analyses: (1) accessibility to health care, (2) evaluation o f health care, {3) perception o f symptoms and threat o f disease, (4) social network characteristics, (5) knowledge about disease, and (6) demographic characteristics. The results o f the study provide a first step in developing a unified framework for explaining health actions. KEY WORDS: health-behavior predictor models; access, psychosocial, and network variables; Smallest Space Analysis.
This research was supported in part by Grant No. HD 00237 from the National Institute of Child Health and Human Development. 1Department of Health Behavior and Health Education, School of Public Health, The University of Michigan, Ann Arbor, Michigan 48109. 123 0160-7715/80/0600-0123 $03.00/0 9 1980 Plenum Publishing Corporation
124
Cummings, Becket, and Maile
INTRODUCTION With the shifting emphasis in health care toward a focus on prevention and early detection of disease, the effectiveness of many health programs is becoming increasingly dependent upon the willingness of individuals to accept a more active role in caring for their health. Despite the fact that the individual stands to benefit the most from the success of health programs, participation in screening, immunization, and other preventive health efforts, as well as rates of individual compliance to prescribed medical therapies, has been relatively low (Barofsky, 1977; Sackett and Snow, 1979). There is no single or simple answer to the question of why some people choose to take an active role in caring for their health, while others do not. The many different efforts of behavioral scientists to both identify and explain the determinants of voluntary health-related behaviors attest to the fact that the underlying problem is multifaceted and complex. M~Kinlay's (1972) review of the literature on the use of health services, for example, identified six approaches for explaining utilization behavior: economic, sociodemographic, geographic, sociopsychological, sociocultural, and organizational. Over the past two decades, a number of theoretical frameworks have appeared which attempt to account for health actions. Most notable, in terms of predictive ability and frequency of citation, have been the models proposed by Andersen (1968), Anderson and Bartkus (1973), Antonovsky and Kats (1970), Fabrega (1973, 1974), Green (1975), Hochbaum (1958), Kar (1977, 1978), Kasl and Cobb (1966a,b) (actually two models, one for "health" and one for "illness" behaviors), Kosa and Robertson (1975), Langlie (1977), Mechanic (1968), Rosenstock (1966) [while the Rosenstock and Hochbaum models were derived from the same body of theory, each, as published, contains variables not found in the other - cf. Kirscht et al. (1978)], and Suchman (1966). Although these 14 models differ considerably in the theoretical perspectives used to explain behavior, in the types of behaviors to be explained, and in the terms employed to label the different dimensions and variables, the general classes of factors included in each of the models appear, at least superficially, to be quite similar. For example, all of these formulations possess one or more variables which represent the individual's evaluation of various health actions (e.g., "treatment benefits," "value of health services," and "advantages of action"). The models advanced by Mechanic, Rosenstock, Langlie, Kasl and Cobb, Fabrega, Hochbaum," Andersen, Anderson and Bartkus, Antonovsky and Kats, and Kosa and Robertson include variables which assess the individual's perception of symptoms in relation to disease threat (e.g., "perceptual salience of symptoms," "assessment of symptoms," "symptom sensitivity," and "perceived susceptibility to illness conditions"). Finally, 10 of the 14 models contain variables which reflect factors that facilitate or inhibit access to health care (e.g., "availability of treatment resources," "monetary costs," "awareness of health
Combining Variables Used to Explain Health Actions
125
facilities," and "distance to health facilities"). The student of the health-behavior literature might therefore reasonably suspect, upon examination of extant models, that the actual number of truly distinct concepts believed to be important in understanding and predicting health-related behaviors is far lower than the more than 100 variables these models have generated. A listing of the variables and a brief summary of each model are presented in Table I. But, while the variables can be combined into a limited number of groupings on a face-validity basis, it might be argued that the creators of the models are aware of more subtle aspects and distinctions which, if made explicit, would require that these like-appearing variables not be viewed as highly comparable. (Such an approach also provides a conservative test of intermodel comparability, since one might expect each model builder to wish to preserve what are felt to represent the unique characteristics and contributions of his/her model to the field.) Thus, any attempt to reduce the great multiplicity of concepts and variables (in order to move toward a more simple, unified set of compatible concepts to explain health-related behaviors) must be based upon data obtained from the various model builders themselves. The purposes of this paper are to explore the selected approaches and models which have been advanced to explain health actions in terms of structural similarities and differences identified by a panel of judges who are relevant experts in this field and to use these expert evaluations to construct a general taxonomy of factors affecting health-related behaviors. The latter objective should be useful for at least two reasons. First, these broader dimensions can provide a more complete framework for making decisions about the types of variables which ought to be included in future research on health behavior. Second, a greater understanding of the structure of a particular class of variables should aid in the development of more reliable and precise measures of the concept they are supposed to represent.
METHOD The data analyzed here are derived from judges' assessments of a complete listing of the variables included in each of the 14 models presented in Table I. Eighteen variables found to have been labeled and defined in exactly the same way in two or more models were dropped from the listing. A total of 109 different variables was retained for use in the study.2 A description of each of the variables is presented in the Appendix. 2Ultimately, 10 of the rated variables were excluded from the analyses because they were classified as "miscellaneous" by five of eight judges; these variables were "competing possible interpretation of symptoms," "residential mobility,.... illness recognition and labeling," "selection of treatment plans," "treatment plans," "response to illness," "classification of manipulative actions," "complexity of behavior," "trialability," and "observability."
Medical care utilization
Choice of alternative health services
Health behavior (dental health)
Anderson and Bartkus
Antonovsky and Kats
Predicted variables
Andersen
Model
The model indicates that sociopsychological factors such as symptom sensitivity and appraisal of health service, need for medical care, and ability to pay for health service directly affect choice of health service. Sociodemographic and ecological factors are seen as influencing utilization indirectly through their influence on intervening sociopsychological variables.
The model hypothesizes that the amount of health services used by a family will be a function of the predisposing (family composition, social structure, and health beliefs), enabling (family and community resources), and need (health status) characteristics of the family. Each component is assumed to make an independent contribution to understanding differences in use of health services.
Notes
Salience of health measured as the individual's willing- The model hypothesizes that behavior is goal ness to spend a lot of money on health action and directed and that three goals can be specithe amount of discussion about health problems fied: (1) enhancement of health or avoidwith friends and relatives. Effective motivation. ance of illness, (2) achievement of approval Knowledge about the disease. Anxiety about taking by others, and (3) achievement of self-apthe health action. Financial difficulty. proval. The three goals are seen as being not mutually exclusive but rather reinforcing of each other. Two other types of factors are seen as influencing the motivation: (1) blockage variables which prevent the behavior from occurring and (2) conditioning variables which intervene between the motivating and the blockage variables.
Appraisal of adequacy of care. Perceptions of friends' appraisal of the adequacy of health services. Perception of medical symptoms. Symptom sensitivity. Need for medical care. Ability to pay for health services. Availability of health service. Regular family physician. Sociodemographic factors.
Family composition, such as age, sex, and family size. Social structure measured by characteristics of the family's main earner, such as employment, social class, educational level, occupation, race, and ethnic affiliation. Health beliefs about medical care, physicians, and disease. Family resources measured by the family's ability to pay for health services. Community resources measured as the availability of health services and health education level in the community. Perception of health status. Response to illness.
Predictors
Table 1. Selected Models of Health-Related Bahviors
~r ~'
Illness behaviors
Adoption of health behaviors
Fabrega
Green
Social pressures. Frequency of contact and communication with individuals of the same social stratum. Social participation. Relative advantages of the action. Contact and communication with persons outside one's own social stratum. Compatability with existing values, past experience, and needs of the individual. Complexity of the action. Trailability. Observability. Sociodemographic characteristics.
Illness recognition and labeling. Illness disvalue. Treatment plans. Assessment of treatment plans. Treatment benefits. Treatment costs. Net benefits or utility. Selection of treatment plans.
The diffusion and adoption model includes three categories of factors which influence the adoption of health practices. These include: (1) environmental or system factors such as socioeconomic stratum and social norms; (2) characteristics of the adopter, such as contact and communication with individuals within and outside one's own social stratum and participation in social groups; (3) characteristics of the innovation such as its relative advantages compared to other things, its compatibility with existing values, past experiences, and needs of the individual, its complexity, its trailability, and its observability to others.
The model includes nine stages of information processing which lead the person to the decision to behave in a certain way. The first two stages correspond to the recognition and evaluation of symptoms. Stage 3 refers to the formulation of treatment plans which is based on the individual's past experience with medical care, advice from friends, and advice from health professionals. Stages 4, 5, 6, and 7 relate to the individual's evaluation of different treatment plans. Stage 8 involves the individual's decision to take action. Finally, stage 9 refers to the processing and evaluation of information obtained in stages 1 - 8 .
~ o
P.
~.
~"
~. 3"
r
Kasl and Cobb
Health behaviors (contraceptive
Kar
Health behaviors
use)
Health behaviors (medical screening)
Predicted variables
Hochbaum
Model
I m p o r t a n c e of health, perceived susceptibility to and severity of a health condition constitute the threat of disease c o m p o n e n t . Factual i n f o r m a t i o n a b o u t the disease and health action, social characteristics. Past utilization of medical service. Costs o f action vs. inaction. Perceived probability that action produces desired outcome. Perceived efficacy of health action.
Sociocultural determinants, includes social norms, interpersonal influence and c o m m u n i c a t i o n . Situational d e t e r m i n a n t s include knowledge or awareness of service, costs, and p r o x i m i t y to service. Sociopsychological d e t e r m i n a n t s include include intentions, personal aspirations, values, and prior experience with health service. Sociostructural determinants include social status and demographics.
Belief in the possibility of contracting disease. Belief in the benefits o f health action. Features of the source o f care such as cost and location o f service, and the individual's opinion regarding the quality of care. Social forces. Appearance of symptoms.
Predictors
Table I. C o n t i n u e d
The m o d e l h y p o t h e s i z e s that behavior undertaken in t h e absence o f s y m p t o m s is influenced directly by threat of disease and beliefs concerning the value of the health action. Social characteristics and knowledge are seen as influencing b o t h the perceived threat of disease and the value of action and t h u s indirectly influencing behavior. Past utilization o f medical services, t h e ratio o f the costs of action to the
T h e m o d e l holds that when all four categories o f d e t e r m i n a n t s are favorable, t h e behavior will take place. If certain factors are n o t favorable, the behavior m a y or m a y n o t occur, depending on the strength o f the factors inhibiting the behavior. Moreover, it is a s s u m e d that all four categories of d e t e r m i n a n t s are independent.
Behavior is viewed as a f u n c t i o n o f three sets of factors: (1) the individual's psychological state o f readiness which includes beliefs a b o u t susceptibility to disease and beliefs a b o u t the benefits of taking a health action; (2) situational influences which include appearance of sympt o m s which are interpreted as relating to the disease and influences exerted by other people toward or away f r o m the idea of taking t h e h e a t h action; and (3) envir o n m e n t a l condition which includes availability and access to health services.
Notes
~~"
f-
